Machine Learning Analysis for Side-Channel Attacks over Elliptic Curve Cryptography

Abstract

This paper presents experimental evaluation on various Machine Learning (ML) models used for Simple Power Attacks (SPA) on a device which uses an Elliptic Curve Cryptography (ECC). The device used in this investigation is an Atmega 328 U which has an 8-bit architecture. The paper provides a comparison between the detection of the secret key using ML algorithms. Based on the results obtained in the research, certain recommendations are made.